A pipeline, after the first three clients

The first three came from people you knew

That route is covered in the module on getting seen, and it works. It also runs out. What follows is how a repeatable flow of work is built once you have exhausted the people who already trust you.

Say what you do in one sentence

“Freelance data consultant” is unmemorable and unforwardable. Nobody hears it and thinks of a specific person with a specific problem.

The sentence that works names a buyer, a problem and an outcome:

“I fix broken reporting for e-commerce companies with under fifty staff: usually a warehouse, dbt models and a dashboard people actually trust, in about six weeks.”

That is narrower than feels comfortable, which is precisely why it works. Somebody who hears it can forward you to the right person without thinking. You can still take other work; the sentence is a door, not a fence.

The sources, ranked by how well they convert

1. Past colleagues and employers. Highest conversion of anything, by a distance. People who have seen you work do not need convincing. Tell every former colleague what you now do, once, without asking for anything.

2. Referrals from clients. Ask at the end of a good project, specifically: ”Do you know anyone else dealing with the reporting problem we just fixed?” A general “let me know if you hear of anything” produces nothing.

3. Subcontracting from agencies and other freelancers. The most underrated source in this lesson. Agencies win more work than they can staff and regularly need a specialist for six weeks. You get steady work with no sales effort at a rate 30% to 50% below what they bill, which is a fair trade for them carrying the client relationship and the payment risk. Approach ten small agencies in your specialism and describe exactly what you can take off their hands.

4. Inbound from public work. Slow, compounding, and the highest quality when it arrives, because somebody who found you through a write-up has already decided you know the subject. This is the freelancing payoff of the writing-in-public lesson.

5. Local business networks. Chambers of commerce, industry associations, trade groups. Unglamorous and full of companies with real problems and no data person.

6. Platforms. Useful, with the caveats below.

7. Cold outreach. Works when specific. A message naming something you noticed about their business, one concrete thing you would do, and no attachment. Ten thoughtful messages beat two hundred templated ones and take about the same time.

Platforms, with the numbers

Fee structures change, so verify before relying on any figure, but the shape is stable and it matters:

• Upwork charges freelancers a service fee on earnings, plus fees on the client side that reduce what clients are willing to offer.
• Fiverr takes a notably larger share, around a fifth.
• Toptal and similar curated networks screen hard, pay meaningfully better, and take their margin on the client side.
• Contra, PeoplePerHour and regional platforms sit in between.

Two honest things about them. The low end is a genuine race to the bottom, where you compete on price with people whose cost of living is lower than yours, and no amount of profile optimisation wins that race. And you do not own the relationship: the platform does, the terms usually forbid taking clients off it, and your income depends on an account that can be suspended.

The sensible use: platforms for the first handful of paid references when you have none, then a deliberate move to direct clients. Treat them as a starting ramp, not a business.

The weekly habit

Half a day a week, permanently:

• Two specific outreach messages.
• One reconnection with a past client or colleague, with no ask attached.
• One piece of public work: a write-up, an answer, a small tool.
• Update the pipeline list: who you spoke to, what you agreed, when you will follow up.

Follow-up is where most freelance work is actually won. A polite second message a week later converts a surprising share of silences, and almost nobody sends it.

The limitation

Selling is the job. Technical people consistently underestimate this and dislike it, and there is no arrangement in which good work markets itself early on. The first ten clients are much harder than the next fifty, the effort is front-loaded and mostly invisible, and this is the part where people quit, usually in month four, about three weeks before the pipeline they built in month two would have started producing.